Databases
StackOverflow2013
Postsdbo
POAtomicity of `write(2)` to a local filesystem
 

Note that there are some explanatory texts on larger screens.

plurals
  1. POAtomicity of `write(2)` to a local filesystem
    primarykey
    data
    text
    <p>Apparently POSIX states that</p> <blockquote> <p>Either a file descriptor or a stream is called a "handle" on the open file description to which it refers; an open file description may have several handles. […] All activity by the application affecting the file offset on the first handle shall be suspended until it again becomes the active file handle. […] The handles need not be in the same process for these rules to apply. -- <a href="http://pubs.opengroup.org/onlinepubs/9699919799/functions/V2_chap02.html#tag_15_05_01" rel="noreferrer" title="2.5.1 Interaction of File Descriptors and Standard I/O Streams">POSIX.1-2008</a></p> </blockquote> <p>and</p> <blockquote> <p>If two threads each call [the write() function], each call shall either see all of the specified effects of the other call, or none of them. -- <a href="http://pubs.opengroup.org/onlinepubs/9699919799/functions/V2_chap02.html#tag_15_09_07" rel="noreferrer" title="2.9.7 Thread Interactions with Regular File Operations">POSIX.1-2008</a></p> </blockquote> <p>My understanding of this is that when the first process issues a <code>write(handle, data1, size1)</code> and the second process issues <code>write(handle, data2, size2)</code>, the writes can occur in any order but the <code>data1</code> and <code>data2</code> <em>must</em> be both pristine and contiguous.</p> <p>But running the following code gives me unexpected results.</p> <pre class="lang-c prettyprint-override"><code>#include &lt;errno.h&gt; #include &lt;stdio.h&gt; #include &lt;stdlib.h&gt; #include &lt;string.h&gt; #include &lt;fcntl.h&gt; #include &lt;unistd.h&gt; #include &lt;sys/wait.h&gt; die(char *s) { perror(s); abort(); } main() { unsigned char buffer[3]; char *filename = "/tmp/atomic-write.log"; int fd, i, j; pid_t pid; unlink(filename); /* XXX Adding O_APPEND to the flags cures it. Why? */ fd = open(filename, O_CREAT|O_WRONLY/*|O_APPEND*/, 0644); if (fd &lt; 0) die("open failed"); for (i = 0; i &lt; 10; i++) { pid = fork(); if (pid &lt; 0) die("fork failed"); else if (! pid) { j = 3 + i % (sizeof(buffer) - 2); memset(buffer, i % 26 + 'A', sizeof(buffer)); buffer[0] = '-'; buffer[j - 1] = '\n'; for (i = 0; i &lt; 1000; i++) if (write(fd, buffer, j) != j) die("write failed"); exit(0); } } while (wait(NULL) != -1) /* NOOP */; exit(0); } </code></pre> <p>I tried running this on Linux and Mac OS X 10.7.4 and using <code>grep -a '^[^-]\|^..*-' /tmp/atomic-write.log</code> shows that some writes are not contiguous or overlap (Linux) or plain corrupted (Mac OS X).</p> <p>Adding the flag <code>O_APPEND</code> in the <code>open(2)</code> call fixes this problem. Nice, but I do not understand why. POSIX says</p> <blockquote> <p>O_APPEND If set, the file offset shall be set to the end of the file prior to each write.</p> </blockquote> <p>but this is not the problem here. My sample program never does <code>lseek(2)</code> but share the same file description and thus same file offset.</p> <p>I have already read similar questions on Stackoverflow but they still do not fully answer my question.</p> <p><a href="https://stackoverflow.com/questions/6896011/atomic-write-on-file-from-two-process">Atomic write on file from two process</a> does not specifically address the case where the processes share the same <em>file description</em> (as opposed to the same file).</p> <p><a href="https://stackoverflow.com/questions/7660293/how-does-one-programmatically-determine-if-write-system-call-is-atomic-on-a-pa">How does one programmatically determine if “write” system call is atomic on a particular file?</a> says that</p> <blockquote> <p>The <code>write</code> call as defined in POSIX has no atomicity guarantee at all.</p> </blockquote> <p>But as <a href="http://pubs.opengroup.org/onlinepubs/9699919799/functions/V2_chap02.html#tag_15_09_07" rel="noreferrer" title="2.9.7 Thread Interactions with Regular File Operations">cited above</a> it does have some. And what’s more, <code>O_APPEND</code> seems to trigger this atomicity guarantee although it seems to me that this guarantee should be present even without <code>O_APPEND</code>.</p> <p>Can you explain further this behaviour ?</p>
    singulars
    1. PO
      singulars
      1. PTAnswer
    1. This table or related slice is empty.
    1. PTQuestion
    1. USCommunity
    1. USkmkaplan
    plurals
    1. PL
      singulars
      1. LTLinked
    2. PL
      singulars
      1. LTLinked
    1. PL
      singulars
      1. LTLinked
    1. This table or related slice is empty.
    1. PO
      singulars
      1. PTAnswer
    2. PO
      singulars
      1. PTAnswer
    3. PO
      singulars
      1. PTAnswer
    1. VO
      singulars
      1. POAtomicity of `write(2)` to a local filesystem
      1. This table or related slice is empty.
      1. VTUpMod
    2. VO
      singulars
      1. POAtomicity of `write(2)` to a local filesystem
      1. This table or related slice is empty.
      1. VTUpMod
    3. VO
      singulars
      1. POAtomicity of `write(2)` to a local filesystem
      1. USosgx
      1. VTFavorite
    1. CODoes OSX claim POSIX08 conformance? I don't think so. (I believe they claim '03 compliance only.)
      singulars
      1. POAtomicity of `write(2)` to a local filesystem
      1. USDavid Schwartz
    2. COGood point, according to http://images.apple.com/macosx/docs/OSX_for_UNIX_Users_TB_July2011.pdf it is “Open Brand UNIX 03”. I’ll have to check out what that means.
      singulars
      1. POAtomicity of `write(2)` to a local filesystem
      1. USkmkaplan
    3. COA lot of people will answer based on pre-'08 rules, where write was only atomic on pipes and even then only under certain conditions. A lot of platforms still don't support the '08 semantics. And a lot of platforms that claim to, still have one or more filesystems that don't.
      singulars
      1. POAtomicity of `write(2)` to a local filesystem
      1. USDavid Schwartz
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload