<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
I think the issue is the kernel is extremely concerned with the<br>
efficiency of the syscall path.<br>
<br>
Very legitimately some benchmarks just measure that one path to see<br>
how many thousands of syscalls per second can be made.<br>
<br>
To accelerate that path as much as possible, the linux kenel chooses<br>
not to incur the overhead of preserving the FP registers on every<br>
syscall.<br>
<br>
So kernel code that uses FP must first ensure any registers it uses<br>
are preserved. I don't recall ever writing any FP kernel code, so I<br>
don't know what facilities are available to do that.<br>
<font color="#888888"><br>
Greg<br></font></blockquote><div><br>Greg<br><br>I had a vague idea about overhead being incurred due to the mode switch, but <br>your clear explanation makes it a lot easier to understand _where_ this happens. <br>It's very helpful, thank you for your response.<br>
<br>Julie<br></div></div><br>