<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns="http://www.w3.org/TR/REC-html40">
<head>
<meta http-equiv=Content-Type content="text/html; charset=us-ascii">
<meta name=Generator content="Microsoft Word 11 (filtered medium)">
<style>
<!--
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:"Times New Roman";}
a:link, span.MsoHyperlink
{color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{color:purple;
text-decoration:underline;}
pre
{margin:0in;
margin-bottom:.0001pt;
font-size:10.0pt;
font-family:"Courier New";}
span.EmailStyle17
{mso-style-type:personal-compose;
font-family:Arial;
color:windowtext;}
@page Section1
{size:8.5in 11.0in;
margin:1.0in 1.25in 1.0in 1.25in;}
div.Section1
{page:Section1;}
-->
</style>
</head>
<body lang=EN-US link=blue vlink=purple>
<div class=Section1>
<p class=MsoNormal><font size=3 face="Times New Roman"><span style='font-size:
12.0pt'>Roland Dreier wrote: <o:p></o:p></span></font></p>
<pre style='margin-bottom:12.0pt' wrap=""><font size=2 face="Courier New"><span
style='font-size:10.0pt'> Sean> I think that the issue that Arlin is hitting is that once he<br>
Sean> calls blah_get_event() he doesn't have an easy way to<br>
Sean> release the thread. This may turn out to be a uCM / uAT<br>
Sean> issue, and not a verbs issue.<br>
<br>
I thought about this some more. I think the real issue is probably<br>
that the app is sleeping blah_get_event() on a blocking fd. If the<br>
app doesn't want to block in that function forever, then it should use<br>
poll() with a wake-up pipe() fd as well as the event fd. (I'm also<br>
not sure what's so bad about pthread_kill() as a wake-up)<br>
<br>
- R.<o:p></o:p></span></font></pre>
<p class=MsoNormal><font size=3 face="Times New Roman"><span style='font-size:
12.0pt'>The application (udapl completion thread) is polling on the FD so it is
very easy to add a pipe() fd to wakeup. I was just expecting the close_fd in
ibv_close_device() to wakeup the poll (or blocking get_event) so I could exit
the thread without a pipe or a kill. <br>
<br>
pthread_kill requires me to add a sig_handler to catch the signal and since
uDAPL is a library, there is a chance of uDAPL replacing a signal handler already
setup by the application for the process group. <br>
<br>
-arlin</span></font><font size=2 face=Arial><span style='font-size:10.0pt;
font-family:Arial'><o:p></o:p></span></font></p>
</div>
</body>
</html>