You'll get bus error when you access unaligned pointers. like when you access an integer pointer which is not 4 byte aligned. And gcc3 on sparc is very buggy, you should really disable all optimization.<br><br><br><br><div>
<span class="gmail_quote">On 10/6/06, <b class="gmail_sendername">Christian Bienia</b> <<a href="mailto:cbienia@cs.princeton.edu">cbienia@cs.princeton.edu</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Hi,<br><br>despite all my efforts, I haven't been able to find the bug so far. :-(<br>But it might be that gcc is generating incorrect Sparc code. Here's what<br>I've tried:<br><br>- Blastwave's gcc4: bug always shows up.
<br>- Sun's preinstalled gcc3: bug only shows up if the optimizer is<br>enabled. x264 works fine if compiled with -O0.<br>- CoolTools gcc (gcc4 frontend + Sun's cc backend): x264 works fine (!)<br><br>Depending on the optimization level (x264 uses -O1 as debug default, I
<br>also tried -O0), the segfault occurs at different places in the code,<br>but always somewhere in assembly code for which the compiler cannot show<br>any C code (such as the PLT case I described in my earlier email).<br>
<br>I suspect that the gcc Sparc backend generates incorrect code. In any<br>case, only p8x8 partitions seem to be affected, and x264 runs without<br>problems if they're deactivated.<br><br>What do you guys think about that?
<br><br>- Chris<br><br>--<br>This is the x264-devel mailing-list<br>To unsubscribe, go to: <a href="http://developers.videolan.org/lists.html">http://developers.videolan.org/lists.html</a><br><br></blockquote></div><br>