kopia lustrzana https://gitlab.com/sane-project/website
94 wiersze
4.2 KiB
HTML
94 wiersze
4.2 KiB
HTML
<!-- received="Mon Jul 5 19:08:38 1999 PDT" -->
|
||
<!-- sent="Mon, 05 Jul 1999 22:13:30 -0400" -->
|
||
<!-- name="Douglas Gilbert" -->
|
||
<!-- email="dgilbert@interlog.com" -->
|
||
<!-- subject="Re: mustek aha1520B problems" -->
|
||
<!-- id="" -->
|
||
<!-- inreplyto="mustek aha1520B problems" -->
|
||
<title>sane-devel: Re: mustek aha1520B problems</title>
|
||
<h1>Re: mustek aha1520B problems</h1>
|
||
<b>Douglas Gilbert</b> (<a href="mailto:dgilbert@interlog.com"><i>dgilbert@interlog.com</i></a>)<br>
|
||
<i>Mon, 05 Jul 1999 22:13:30 -0400</i>
|
||
<p>
|
||
<ul>
|
||
<li> <b>Messages sorted by:</b> <a href="date.html#38">[ date ]</a><a href="index.html#38">[ thread ]</a><a href="subject.html#38">[ subject ]</a><a href="author.html#38">[ author ]</a>
|
||
<!-- next="start" -->
|
||
<li> <b>Next message:</b> <a href="0039.html">Brian B.: "once more on the aha152x problem"</a>
|
||
<li> <b>Previous message:</b> <a href="0037.html">Juergen Scherer: "Microtek V600"</a>
|
||
<li> <b>Maybe in reply to:</b> <a href="0006.html">aanderson =: "mustek aha1520B problems"</a>
|
||
<!-- nextthread="start" -->
|
||
<!-- reply="end" -->
|
||
</ul>
|
||
<!-- body="start" -->
|
||
"aanderson = (A.N. Anderson - Idaho Research Software)" wrote:<br>
|
||
<i>> </i><br>
|
||
<i>> >Douglas Gilbert <<a href="mailto:dgilbert@interlog.com">dgilbert@interlog.com</a>> responded:</i><br>
|
||
<i>> ></i><br>
|
||
<i>> >Did the bigger buffers improve performance and lessen</i><br>
|
||
<i>> >the following problems?</i><br>
|
||
<i>> ></i><br>
|
||
<i>> </i><br>
|
||
<i>> To be honest I didn't see any difference, which is why I think</i><br>
|
||
<i>> think there are problems with the driver. The performance wasnt</i><br>
|
||
<i>> bad in either case, tho, except for the hangups and crashes, the</i><br>
|
||
<i>> latter of which typically occurred on large scans where improvement</i><br>
|
||
<i>> might have been more apparent. I was careful to get the buffer logistics</i><br>
|
||
<i>> right, altho I am suspicious that it seemed to crash more often with</i><br>
|
||
<i>> the large buffers..</i><br>
|
||
<p>
|
||
Yes with the old driver bigger buffers<br>
|
||
increased the probability of not being able to<br>
|
||
find memory which led to an oops.<br>
|
||
<p>
|
||
[snip]<br>
|
||
<i>> ></i><br>
|
||
<i>> >Time taken to DMA/block_move data to and from the SCSI device</i><br>
|
||
<i>> >and any polling that the aha152x driver does will _not_</i><br>
|
||
<i>> >show on a command prefixed with "time ".</i><br>
|
||
<i>> ></i><br>
|
||
<i>> I assumed the polling should show up under 'sys', but that there</i><br>
|
||
<i>> should still be lots of idle time between scsi commands.</i><br>
|
||
<i>> What concerned me more was that it seemed to keep on polling even after</i><br>
|
||
<i>> the scan stopped.</i><br>
|
||
<p>
|
||
I should explain this a little more in reference to<br>
|
||
the Linux sg driver. If data is being read from <br>
|
||
the scanner it is first transferred to "dma"<br>
|
||
kernel buffers then it is transferred from there to<br>
|
||
the user buffer. This double handling is why this<br>
|
||
approach is sometimes called indirect IO. The time<br>
|
||
to do the first transfer is not registered against<br>
|
||
the user (and is often done under interrupt<br>
|
||
privilege). The second transfer (into the user space)<br>
|
||
will show up as 'sys' time as you predicted.<br>
|
||
<p>
|
||
Before people jump on me about the double handling,<br>
|
||
it is not easy to implement direct IO under Linux<br>
|
||
and only one driver that I am aware of does it: bttv.<br>
|
||
The memory that you allocate for the direct DMA<br>
|
||
transfer needs to be locked down while the transfer<br>
|
||
is pending, this means not swapping out that part<br>
|
||
of the app and not allowing it to be killed<br>
|
||
amongst other things. Besides, Linus doesn't like<br>
|
||
direct IO.<br>
|
||
<p>
|
||
[snip]<br>
|
||
<p>
|
||
Doug Gilbert<br>
|
||
<p>
|
||
<pre>
|
||
--
|
||
Source code, list archive, and docs: <a href="http://www.mostang.com/sane/">http://www.mostang.com/sane/</a>
|
||
To unsubscribe: echo unsubscribe sane-devel | mail <a href="mailto:majordomo@mostang.com">majordomo@mostang.com</a>
|
||
</pre>
|
||
<!-- body="end" -->
|
||
<p>
|
||
<ul>
|
||
<!-- next="start" -->
|
||
<li> <b>Next message:</b> <a href="0039.html">Brian B.: "once more on the aha152x problem"</a>
|
||
<li> <b>Previous message:</b> <a href="0037.html">Juergen Scherer: "Microtek V600"</a>
|
||
<li> <b>Maybe in reply to:</b> <a href="0006.html">aanderson =: "mustek aha1520B problems"</a>
|
||
<!-- nextthread="start" -->
|
||
<!-- reply="end" -->
|
||
</ul>
|